The Non-Terrestrial Networks (NTNs) have received significant attention in the global market for their contribution to the low-altitude economy. As a critical component, Low-Earth Orbit (LEO) satellite networks facilitates information exchange but also poses security risks due to its broadcast nature. To this end, this paper presents a novel cross-chain data transmission framework based on satellite off-chain channels, integrating the security features of blockchain with the high-efficiency transmission capabilities of off-chain mechanisms to enhance the data delivery performance. Within this framework, we propose a multi-dimensional resource allocation model that incorporates key constraints, such as communication windows, orbital dynamics, and chain bandwidth limitations. A real-time window scheduling strategy is further developed to address both latency sensitivity and data priority. To facilitate efficient optimization under complex constraints, we propose the Hypergeometric Dynamic Phase Optimization (HDPO) algorithm. HDPO attains robust convergence by leveraging hypergeometric functions, enhances adaptive exploration via dynamic phase adjustment, and mitigates both local optima entrapment and parameter sensitivity through hybrid search strategies. Experimental evaluations demonstrate that the proposed on-chain/off-chain collaborative transmission scheme significantly reduces transmission delay and improves throughput compared to traditional on-chain methods. Moreover, the HDPO algorithm achieves the highest data transmission success rates under various interference levels, highlighting its superior robustness and environmental adaptability.

With the development of Sixth Generation (6G) Internet of Things (IoT) technology, ensuring data reliability and security in networks has become a critical issue. To address the identification of abnormal behaviors in network traffic, this study proposes an anomaly traffic detection algorithm combining wavelet analysis and machine learning. By utilizing wavelet analysis, this paper ex-tracts time-frequency features from Fifth Generation (5G) core network traffic data, which effectively capture abrupt changes and periodic fluctuations in the data. Combining deep learning models, particularly dynamic convolution and attention mechanisms, this method adaptively optimizes the feature extraction process, enhancing the model’s sensitivity and accuracy in detecting key traffic features. Experimental results demonstrate that the proposed algorithm outperforms traditional methods in multiple standard datasets, with superior performance in accuracy, precision, recall, and other evaluation metrics.

Modern transportation networks, with their complexity and dynamic nature, have a substantial demand for intelligent vehicles. Developing effective production strategies for smart vehicles is essential to reducing both production costs and energy consumption. Traditional vehicle production planning has largely depended on heuristic algorithms and solvers, which lack scalability and are susceptible to local optima. Furthermore, existing solutions do not concurrently address both dynamic and regular vehicle production planning. To overcome these limitations, this paper proposes an effective optimizing method for large-scale smart manufacturing within intelligent transportation networks using Federated Reinforcement Learning. In our proposal, the Gated Recurrent Unit and Asynchronous Advantage Actor Critic (A3C) reinforcement algorithms are employed to develop a Dynamic Optimizing Planning Module(DOPM), which can output an excellent solution of 1000 vehicles within 5 seconds. A High-Quality Processing Module(HQPM) is constructed by the Transformer with A3C, significantly enhancing the production plan’s quality. Finally, the proposed methods will integrate with Federated Learning (FL) to establish a scalable, privacy-preserving intelligent manufacturing scheduling framework for transportation networks. Experimental results demonstrate that our work significantly outperforms traditional solutions, achieving over a 93% improvement in solving speed and reducing constraint violations by more than 95%.

Abstract Recently, face recognition based on homomorphic encryption for privacy preservation has garnered significant attention. However, there are two major challenges with homomorphic encryption methods: the security and efficiency of face recognition systems. We present a more efficient and secure PUM (Privacy preserving security Using Multi‐key homomorphic encryption) mechanism for facial recognition. By integrating feature grouping with parallel computing, we enhance the efficiency of homomorphic operations. The use of multi‐key encryption ensures the security of the facial recognition system. This approach improves the security and speed of facial recognition systems in cloud computing scenarios, increasing the original 128‐bit security to a maximum of 1664‐bit security. In terms of efficiency, comparing encrypted images takes only 0.302 s, with an accuracy rate of 99.425%. When applied to a campus scenario, the average search time for a facial template library containing 700 encrypted features is approximately 1.5 s. Consequently, our solution not only ensures user privacy but also demonstrates superior operational efficiency and practical value. In comparison to recently emerged ciphertext facial recognition systems, our solution has demonstrated notable enhancements in both security and time efficiency.

Vehicular ad-hoc networks (VANETs) have become the predominant technology in the current era. Although VANETs have numerous benefits, they are prone to different types of attacks owing to their open nature. Therefore, security plays a crucial role in VANET systems. Ensuring a safe and dependable vehicular communication system is crucial when performing anonymous authentication and group key agreement. Many related works have been proposed based on signature aggregation and group key management; however, they suffer from high computational and communication costs. Hence, in this work, signature aggregation scheme is proposed in such a way that the computational overhead is significantly reduced. Moreover, an ECC-based group management scheme is proposed to secure group communication. In comparison to recent works, the proposed work generates and verifies signatures with an efficiency of 50.03% and 26.26%, respectively. Furthermore, 72.32% and 35.45% efficient in terms of transmission overhead and serving ratio when compared to recent works. To validate this work practically, printed antenna consisting of four elements arranged in a linear array is developed for the intended use. Security analysis is performed in formal and informal ways to prove the security strength of the proposed method. Finally, the performance is validated with similar works using the Cygwin platform with the PBC library.

Beyond 5G-enabled edge computing networking (ECN) will further deploy computing and communication resources to the edge of the networks. Then, edge service demands for Internet of Things (IoT) applications are becoming more and more diverse, while the corresponding routing service capability is limited and not flexible enough to deal with the demands of ECN, which then leads to reducing the inherent routing capability of ECN. It becomes extremely difficult for ECN to support diversified demands and provide diverse IoT applications quickly and flexibly. In this article, we propose a novel and customized deep routing mechanism for IoT applications in ECN, in which the network slicing and deep learning methods are jointly applied and leveraged. First, we design a new ECN architecture that formulates four kinds of network slices to cope with various IoT scenarios, which are eMBB, uRLLC, mMTTC, and backup slices. Second, using these slices, we can customize the ECN environment flexibly, based on which we propose the corresponding routing method for the purpose of fast and efficient service delivery. In particular, the mapping between network slices and the infrastructure is established with the object of maximizing the resource utilization. Then, the routing is designed and customized by using the deep learning model. Lastly, the experimental results show that the deep customized mechanism designed in this article can reduce the average loss rate of the model, decrease the average delay, as well as improve the average resource utilization compared with the existing studies.

Search and rescue (SAR) assumes primary focus during the post-disaster response phase. In recent years, the rapid advances in autonomous aerial vehicle (AAV) target detection technology have opened up new possibilities for SAR operations. However, the images captured by AAV exhibit considerable variation as they dynamically operate at different altitudes, posing challenges for rescue teams in identifying inconspicuous rescue targets. Furthermore, rescuers can hardly trust a detection model with an opaque decision-making process. To this end, we propose an explainable Region of Interest (RoI) attention-based AAV target detection network (RAXNet) capable of detecting small rescue targets with visual explanations. In this model, we first adopt path aggregation network (PANet) as the neck to extract features from different scales. Then, a RoIAttention module is designed to enhance the small target features while providing visual explanations. Specifically, we employ the RoIAlign and nonmaximum suppression methods to obtain region proposals of small targets in the low-level feature layer, followed by an attention-based feature enhancer to focus the extracted region proposals. By combining the enhanced features with the original ones, RAXNet can detect inconspicuous rescue targets and offer corresponding visual explanations, improving model credibility and facilitating rescue efficiency. Finally, we build an AAV visualization system to help rescuers assess disaster sites in real time. The effectiveness and explainability of the proposed method is demonstrated on the VisDrone DET benchmark and RescueNet datasets.

With the explosive growth in Industrial Internet of Things (IIoT) devices, the volume of multimedia data in the field of flexible manufacturing has also increased significantly in recent years, especially the vast amount of unlabelled image data. Image annotation provides machines with a more natural way to interact with users, enhancing the level of intelligence in IIoT flexible manufacturing. This article proposes a multifeature fusion multikernel learning image annotation method to tackle imbalanced label distribution, image weak labeling, and varying representational abilities of features. Initially, oversampling techniques with synthetic minority class samples address the influence of minority classes, while a label enhancer extends label vectors to overcome the influence of weak labeling. Subsequently, the integration of traditional visual features with deep features based on multikernel learning is investigated to enhance feature representation capability. This approach combines complementary information from multiple features, establishing intrinsic connections between images and annotated keywords. Experimental evaluations are conducted on three benchmark datasets, comparing our method with several classical methods. Evaluation results demonstrate that our proposed method captures semantic information more accurately and comprehensively. By effectively accomplishing automatic image annotation, our method can enhance human-machine-interaction to improve the level of intelligence in IIoT flexible manufacturing.

The integration of Cognitive Internet of Things (IoT) sensors with autonomous aerial vehicles (AAVs) has transformed industrial sectors, such as monitoring, logistics, and infrastructure inspection. However, the advancement of visual synthesis technologies like generative adversarial networks and diffusion models has introduced significant risks by enabling the creation of highly realistic AI-manipulated content, making the detection of falsified imagery increasingly challenging. Existing detection methods, largely based on convolutional neural networks (CNNs), focus primarily on global image features and often overlook crucial relational connections, limiting their robustness and generalization. To overcome these limitations, we propose a novel dual-stream architecture that integrates global feature extraction with relational feature learning. By combining the CLIP model with a graph-based topology, our approach identifies hard-to-detect samples and processes them through a graph convolutional network (GCN) to capture both structural and relational information. Extensive evaluations validate the robustness and generalization ability of our method across various generative models and real-world perturbations. This approach offers a scalable and reliable solution to ensure data integrity in industrial IoT systems, helping to preserve societal trust in AI-driven applications.

In response to the critical need for rapid post-disaster assessments, this article introduces an innovative application of artificial intelligence (AI) in unmanned aerial vehicles (UAVs) for disaster relief. A lightweight distributed learning algorithm (namely, YO-FR), is designed to enable multiple UAV agents to share and process environmental data, highlighting the importance of data and knowledge-empowered distributed learning. Moreover, we create a real-world mini-data set collected by UAVs for post-disaster road defects (mini-UPRDs), followed by a data enhancement technology to facilitate feature extraction and promote knowledge-driven learning. The viability of YO-FR is underscored by its enhanced detection precision and processing speed, as evidenced by its performance on the enhanced mini-UPRD data set, surpassing that of existing algorithms. By implementing AI algorithms on UAV platforms, this research offers a theoretical and practical foundation for the practical deployment of IUA in critical application areas, such as emergency management and disaster response.

Existing deep learning-based circuit design methods mostly focused on the primary matching of the model itself or circuit data, lacking generalizability and ignoring deep representation of electromagnetic coupling effects in coupled circuits. Therefore, it exhibits difficulties to further improve the accuracy in circuit performance prediction and requires large training datasets. To address these challenges, this article proposes an intelligent circuit design methodology with graph transfer learning (CircuitGTL). Specifically, it achieves the weighted graph modeling of complex electromagnetic environment circuits, where nodes represent components, edges represent the electromagnetic coupling effect between components, and edge weights signify the differential strength of electromagnetic coupling. Hereby, a fused graph representation model, integrating graph isomorphic network and electromagnetic coupling effect-based graph attention network, is proposed to achieve deep representation learning of graphic circuit data. Then, a model- and data-driven graph transfer learning mechanism considering joint optimizing of circuit’s performance matrix and nonperformance indicators is proposed. This is to achieve lightweight cross-electromagnetic topologies parameters optimization. Taking Terahertz (THz) resonant filters as an example to verify the effectiveness of CircuitGTL, numerical results on the MITCircuitGNN experimental dataset show that: compared with state-of-the-art algorithm CircuitGNN, CircuitGTL achieves 9.2% improvement in cross-electromagnetic topologies performance prediction accuracy, 90.9% reduction in model convergence time and 33.6% reduction in the total coverage area of components with only 20% of data requirement; additionally, the design of CircuitGTL has lower-insertion loss, steeper skirts, and higher-passband intersection-over-union. These results provide valuable insights for lightweight, and high-precision design of coupled electromagnetic structures.

Sentiment propagation plays a crucial role in the continuous emergence of social public opinion and network group events. By analyzing the maximum Influence of sentiment propagation, we can gain a better understanding of how network group events arise and evolve. Influence maximization (IM) is a critical fundamental issue in the field of informatics, whose purpose is to identify the collection of individuals and maximize the specific information's influence in real-world social networks, and the sentiments expressed by nodes with the greatest influence can significantly impact the emotions of the entire group. The IM issue has been established to be an NP-hard (nondeterministic polynomial) challenge. Although some methods based on the greedy framework can achieve ideal results, they bring unacceptable computational overhead, while the performance of other methods is unsatisfactory. In this article, we explicate the IM problem and design a local influence evaluation function as the objective function of the IM to estimate the influence spread in the cascade diffusion models. We redefine particle parameters, update rules for IM problems, and introduce learning automata to realize multiple search modes. Then, we propose a multisearch particle Swarm optimization algorithm (MSPSO) to optimize the objective function. This algorithm incorporates a heuristic-based initialization strategy and a local search scheme to expedite MSPSO convergence. Experimental results on five real-world social network datasets consistently demonstrate MSPSO's superior efficiency and performance compared with baseline algorithms.

Modern industrial datacenter networks employ multirooted tree topologies to accommodate a diverse range of cloud applications, which generate heterogeneous traffic with low-latency short flows and high-throughput long flows. Recently, the proposed learning-based load balancing mechanisms are resilient to dynamic network, but they are agnostic to heterogeneous traffic, resulting in large tail delay. In this article, we propose a new deep reinforcement learning (DRL) based load balancing called DRLB, which uses DRL with the distributed distributional deterministic policy gradients algorithm to make (re)routing for long flows, and adopts the weighted cost multipathing mechanism for short flows. Furthermore, this article introduces a traffic feature-based dynamic training cycle mechanism to adaptively adjust the training cycles. The experimental results show DRLB reduces the flow completion time of short flows by up to 58% and improves the throughput of long flows by 38% compared to the state-of-the-art load balancing mechanisms.

In recent years, data-driven remote medical management has received much attention, especially in application of survival time forecasting. By monitoring the physical characteristics indexes of patients, intelligent algorithms can be deployed to implement efficient healthcare management. However, such pure medical data-driven scenes generally lack multimedia information, which brings challenge to analysis tasks. To deal with this issue, this paper introduces the idea of ensemble deep learning to enhance feature representation ability, thus enhancing knowledge discovery in remote healthcare management. Therefore, a multiview deep learning-based efficient medical data management framework for survival time forecasting is proposed in this paper, which is named as "MDL-MDM" for short. Firstly, basic monitoring data for body indexes of patients is encoded, which serves as the data foundation for forecasting tasks. Then, three different neural network models, convolution neural network, graph attention network, and graph convolution network, are selected to build a hybrid computing framework. Their combination can bring a multiview feature learning framework to realize an efficient medical data management framework. In addition, experiments are conducted on a realistic medical dataset about cancer patients in the US. Results show that the proposal can predict survival time with 1% to 2% reduction in prediction error.

Summary In recent years, smart healthcare, artificial intelligence (AI)‐aided diagnostics, and automated surgical robots are just a few of the innovations that have emerged and gained popularity with the advent of Healthcare 4.0. Such technologies are powered by machine learning (ML) and deep learning (DL), which are preferable for disease diagnosis, identifying patterns, prescribing treatments, and forecasting diseases like stroke prediction, cancer prediction and so forth. Nevertheless, much data is needed for AI, ML, and DL‐based systems to train effectively and provide the desired outcomes. Further, it raises concerns about data privacy, security, communication overhead, regulatory compliance and so forth. Federated learning (FL) is a technology that protects data security and privacy by limiting data sharing and utilizing model information of distributed systems to enhance performance. However, existing approaches are traditionally verified on pre‐established datasets that fail to capture real‐life applicability. Therefore, this study proposes an AI‐enabled stroke prediction architecture consisting of FL based on the artificial neural network (ANN) model using data from actual stroke cases. This architecture can be implemented on healthcare‐based wearable devices (WD) for real‐time use as it is effective, precise, and computationally affordable. In order to continuously enhance the performance of the global model, the proposed FL‐based architecture aggregates the optimizer weights of many clients using a fifth‐generation (5G) communication channel. Then, the performance of the proposed FL‐based architecture is studied based on multiple parameters such as accuracy, precision, recall, bit error rate, and spectral noise. It outperforms the traditional approaches regarding accuracy, which is 5% to 10% higher.

As an important part of Internet of Things (IoT), the Internet of Vehicles (IoV) has been widely used in traffic intersection control, automatic driving, intelligent navigation, etc. However, due to the dynamic topology and high mobility, IoV faces the challenge of frequent disconnections, which will lead to deterioration in the performance of data dissemination. Motivated by the above, air-to-ground (A2G) integrated IoV is used to bridge the communication gaps between terrestrial vehicles to achieve efficient information transmissions. This paper investigates the application of low altitude platform stations (LAPSs) and reconfigurable intelligent surface (RIS) in A2G integrated IoV, where multiple relaying LAPSs equipped with RISs are adopted to improve the spatial multiplexing gain and create the smart radio environment. To make full use of the advantages of LAPS-and-RIS enhanced transmissions, we formulate a weighted sum rate (WSR) maximization problem by jointly considering the location, power, and phase shift. To tackle this challenging non-convex problem, we design an iterative optimization scheme, where three optimization variables are processed in turn. Simulation results demonstrate that the proposed WSR maximization scheme can significantly improve the communication performance in comparison with other state-of-the-art schemes and the baseline scheme.

As the number of devices increases dramatically in the Internet of Things (IoT), features of dense deployment of massive devices generate mutual interference in communication overlapping areas, which will impose an imperative challenge on spectrum resource allocation. To handle this challenge, this article proposes a mutual interference-aware throughput enhancement scheme. For the mutual interference among multiple IoT devices, this scheme first builds an interference hypergraph model to quantify the impact of the mutual interference for each device. According to the main goal of the spectrum resource allocation, this article formulates a graph reinforcement learning (GRL) framework, whose action space is multidimensional discrete, and the reward function is designed to enhance the throughput and mitigate the impact of interference. Then, a graph convolutional network-double dueling deep Q-network-based spectrum resource allocation algorithm is developed upon the proposed GRL framework to extract the mutual interference information from the hypergraph model, and then achieve a dynamic resource allocation for massive IoT. Simulation results prove that the proposed GRL algorithm effectively improves the network throughput compared to the comparison algorithms.

In unmanned aerial vehicle (UAV)-enabled Internet of Things (IoT) networks, the communication ranges between densely deployed IoT devices overlap, resulting in wireless resource conflicts between them. Hence, achieving conflict-free resource allocation is a challenging issue that must be urgently addressed for UAV-enabled IoT networks. To tackle this issue, a hypergraph is used to quantify conflicts, and a federated reinforcement learning (RL)-based resource allocation framework is proposed. Specifically, a conflict graph model is developed for UAV-enabled IoT networks with dense deployments. The model is then converted into a conflict hypergraph model using hypergraph and faction theory. Consequently, the conflict avoidance problem of resource allocation can be reformulated as a hypergraph node coloring problem. The problem is formulated as a Markov decision process, which is solved using a deep RL-based approach. Additionally, to distribute the computational workload across the network and alleviate the burden on the central server, we propose the FedAvg dueling double deep Q-network (FedAvg-D3QN). The proposed FedAvg-D3QN is verified through simulation to have advantages in resource reuse rate and throughput compared to baseline approaches.

Image recognition techniques have become the mainstream solution for indoor unmanned aerial vehicle (UAV) localization and navigation due to the absence of global positioning system. However, unlike autonomous vehicles that enjoy the dividends of the "big model" era, UAVs fail to deploy such models due to the hardware limitations. To this end, this paper proposes a compact all multilayer perceptron (MLP) deep neural network structure that offers a paradigm for theory-guided prompt structural compression of large-scale MLP models. First, we propose a gradient-based sensitivity analysis (GB-SA) method. Unlike existing SA methods, GB-SA obtains nodes’ sensitivity indices with gradient information, openning the possibility of efficient SA for large models. We start by integrating GB-SA with MLP and then extend the mode to MLP-Mixer, which is a promising all-MLP deep neural network. By deeply combining GB-SA and MLP-Mixer, SA-MLP-Mixer emerges as a compact model without reducing the model precision. Finally, we evaluate the effectiveness of the proposed model on the benchmark. The experimental results show that SA-MLP-Mixer has an airborne level model scale and accurate localization capability.

Healthcare systems, empowered by the integration of Artificial Intelligence (AI) and Internet of Things networks, are undergoing significant advancements, ushering in a new era of enhanced treatment experiences and improved quality of life. Edge computing plays a pivotal role as an architectural enabler; however, it also presents numerous energy-related challenges spanning sensors, communication, and edge devices. One of the most formidable challenges is the proliferation of complex communication protocols across various devices, including sensors, reconfigurable intelligent surfaces, smart devices, and edge servers, leading to substantial carbon emissions and energy consumption. To address this challenge, this paper introduces a low-carbon, sustainable edge architecture leveraging AI techniques. Specifically, we develop a deep learning-based radio frequency fingerprint access protocol to facilitate real-time and energy-efficient device access between smart devices and edge gateways. Building upon this foundation, we propose a hybrid quantum-classical optimization algorithm to achieve green data transmission at lower layers for artificial intelligence of things healthcare systems. Simulation results demonstrate that our optimized architecture achieves over 99% identification accuracy using a signal dataset of 50GB obtained from real-world smart devices and practical gateways in a real-world environment, all while maintaining energy-efficient data delivery.

With the rapid evolution of wireless communication technology, smart homes have significantly improved the quality of peoples daily lives by taking advantage of the low latency and high transmission rates of 6G communication technology. Users can now conveniently manage the consumer electronics remotely. However, in the pratical smart home scenarios, the users and consumer electronics communicate with each other through an open public channels where charted and uncharted security risks and privacy vulnerabilities exist. To protect users confidential data from malicious interception and modification, diverse authentication protocols have been proposed so far. However, existing protocols often suffer from efficiency issues or vulnerabilities to known attacks. To address these challenges, this paper proposes a novel three-factor ECC-based anonymous authentication protocol. The protocols security properties can be rigorously proven using the formal analysis under the ROR model. Subsequently, the resistance to numerous types of attacks, including man-in-the-middle and replay attacks, can be demonstrated through informal analysis and the AVISPA verification process. Finally, the protocol is compared with the state-of-the-arts and the results show that the protocol strikes a good balance between security and efficiency and is well suited for smart home environments.

As a typical privacy-aware machine learning paradigm, federated learning (FL) provides facilities to individually train edge clients with their private data and aggregate the central global model. In this way, privacy leakage can be prevented. Massive communication overhead caused by exchanging updated weights between clients and the server is one of the main obstacles in this strategy. Prior work advocates compressing the weights by employing quantization, gradient sparsification, and knowledge distillation approaches. However, most of them cannot be readily applied to secure aggregation in privacy-aware FL. Some research has made great progress in directly utilizing benchmark secure aggregation protocols on top of the non-privacy-aware FL. Graph-based and gradient-based sparsification has been widely adopted in previous studies. However, the results of reducing communication costs are still unsatisfactory. In this paper, we present a novel communication-efficient privacy-ware FL algorithm from a distinct perspective. We design a new Two-Phase Sparsification with Secure Aggregation (TPSSA) algorithm. In the subnetwork phase, we identify sparse subnetworks by freezing the initial random weights in sufficiently overparametrized networks. All edge clients collaboratively train to discover their subnetwork inside a dense randomly weighted neural network. Then the server aggregates to compute the global model. In the gradient phase, for each pair of edge clients, we introduce pairwise multiplicative random masks to identify the sparsification pattern. Then updates from surviving clients can be correctly cancelled out during the aggregation process in the server. Theoretical analysis reveals convergence, privacy and performance guarantee. We show improvements in accuracy, communication, and computation over traditional and sparsified secure aggregation benchmarks on two real-world datasets.

With the rapid integration of Internet of Things (IoT) devices and artificial intelligence (AI) function, the data management and privacy issue has drawn great attentions in intelligent IoT systems where communication infrastructures frequently exchange open data flows over the air. Therefore, lightweight and private access over radio communication pipes becomes a critical but challengeable need for resource-constrained IoT devices due to the limited memory capacity, computing, and energy consumption. In this article, we develop the concept of deep federated scattering fingerprinting aided by differential privacy (DFSF-DP) in which a deep fingerprinting data learning network exploits fingerprinting data to realize lightweight intelligent access and incorporates federated learning with differential privacy to guarantee the data privacy in a way of distributed training. Particularly, first, we employ a wavelet scattering network for the efficient radio frequency fingerprinting (RFF) feature extraction and construct a high information density database. Subsequently, the implementation of distributed learning minimizes the demand for computing resources, by exploiting the full potential of edge and cloud nodes to aggregate the global model. To bolster the data privacy and security, adaptive clipping and gradient noising are incorporated into DFSF-DP. Experimental results demonstrate that DFSF-DP obtains outstanding performance and achieves equivalent advancements while utilizing a mere 25% of the original data set. Moreover, it attains a 93% identification accuracy with 0.1 noise multiplier which confirms the remarkable performance of DFSF-DP while upholding privacy and security considerations.

The rapid advancement of wireless communication and artificial intelligence (AI) has led to a plethora of emerging applications that require exceptional connectivity, minimal latency, and substantial computing resources. The widespread adoption of cloud-edge intelligence is propelling the development of future networks capable of supporting intelligent computing. Mobile edge computing (MEC) technology facilitates the movement of computing resources and storage to the network’s edge, enabling cost-effective offloading of computational tasks for related applications which needs for reduced latency and improved energy efficiency. However, the offloading efficiency is hindered by limitations of wireless transmission capacity. This paper aims to address this issue by integrating reconfigurable intelligent surfaces (RISs) into a cell-free network within an intelligent cloud-edge system. The core idea is to strategically deploy passive RISs around base stations (BSs) to reconstruct the transmission channel and improve the corresponding capacity. Subsequently, we formulate an optimal problem involving joint beamforming for RISs and BSs, which is characterized by non-convexity and complexity. To tackle this challenge, we employ an alternating optimization scheme to ensure the effectiveness of joint beamforming. In particular, deep reinforcement learning (DRL) is leveraged to reduce the computational complexity involved in optimizing task offloading. Additionally, Lyapunov optimization is utilized to model the latency queue and improve the learning efficiency of the offloading framework. We conduct comprehensive evaluations on the wireless system’s capacity, average latency, and energy consumption, considering the integration of RIS with the DRL offloading framework. Experimental results demonstrate that our proposed scheme achieves superior efficiency and robustness.

Federated learning (FL) can train a model collaboratively through multiple remote clients without sharing raw data. The challenge of federated learning (FL) is how to decrease network transmissions. This article aims to reduce network traffic by transmitting fewer neural network parameters. We first investigate similarities of different corresponding layers of convolutional neural network (CNN) models in FL, and find that there is a lot of redundant information in its model feature extractors. For this, we propose a communication-efficient federated aggregation algorithm named FedSL (Federated Split Layers) to reduce the communication overhead. Based on the number of global model layers, the FedSL divides client models into groups in the depth dimension. A Max-Min client selection strategy is employed to select participants for each layer. Each client only transfers partial parameters of those layers that are selected, which reduces the number of parameters. FedSL aggregates the global model in each group and concatenates the parameters of all groups according to the order of layers. The experimental results demonstrate that FedSL improves communication efficiency compared to the algorithms (e.g., FedAvg, FedProx, and MOON), decreasing 42% communication cost with VGG-style CNN and 70% with ResNet-9, while maintaining a similar model accuracy with baseline algorithms.

Wireless powered mobile edge computing (WP-MEC), which combines mobile edge computing (MEC) and wireless power transfer (WPT), is a promising paradigm for coping with the computing power and energy constraints of wireless devices. However, how to realize the online optimal offloading decision and resource allocation in the WP-MEC system is very challenging. This paper studies the system computation completion time (SCCT) minimization problems for WP-MEC networks using non-orthogonal multiple access (NOMA) communication under binary and partial offloading modes. Due to the complexity of the optimization problems and the time-varying nature of the channel state information, we decouple the original problems into a top-problem of optimizing WPT duration and a sub-problem of optimizing resource allocation, and then propose a convolutional deep reinforcement learning online (CDRO) algorithm. For the top-problem, a deep reinforcement learning framework is used to obtain the near-optimal WPT duration, and an incremental exploration policy is designed to balance the exploration accuracy and exploration range to improve the convergence performance of the CDRO algorithm. For the sub-problems, we propose their corresponding low-complexity algorithms based on in-depth analysis and derivation of the optimal offloading decision’s properties. Finally, numerical results show that the proposed CDRO algorithm achieves near-optimal SCCT with low computational complexity, enabling online decision-making in time-varying channel environments.

Intelligent detection techniques for online spamming have been a hot concern in academia. Although much technical progress has been achieved in recent years, two aspects of challenges are still confronted by scholars. For one thing, spamming activities are accompanied by multisource attributes, such as behaviors and semantics. For another, spamming is a cross-platform activity, where multiple platforms are exploited simultaneously to expand the influential reach. The above circumstances actually make spamming detection tend to become a fuzzy detection task. Existing works typically consider one-sided attribute and lack cross-platform multifeature fusion, which limiting the effectiveness of detection. To handle the current challenges, this article proposes a cross-field deep learning-based fuzzy spamming detection approach via the collaboration of behavior modeling and sentiment analysis. First of all, a cross-field deep learning-based technical framework is put forward to implement multisource feature fusion from mixed context. It first extracts multisource features from single fields and then integrates them into a hybrid-field feature space. In addition, three cross-field datasets based on real-world social network datasets are constructed, and utilized in the evaluation of our proposed approach. The findings demonstrate that our proposal improves the detection accuracy by about 7% to 12%, in comparison to five other baseline approaches.

As The cornerstone of human health, physical and mental well-being are intricately linked, influencing both an individual's physical condition and their emotional state [1]. Chronic diseases such as hypertension and diabetes can have a significant impact on mental health, leading to anxiety and depression [2]. Similarly, psychological problems such as stress, anxiety, and depression can weaken the immune system, making individuals more susceptible to physical illnesses. In recent years, the rapid development of technology has brought exciting new possibilities to the field of physical and psychological health. The Internet of Things (IoT) and artificial intelligence (AI) have shown great potential in building a comprehensive health management system that empowers individuals to take a more proactive role in their well-being.

The expanding number of low cost sensors and smart devices drives the internet-of-things (IoT) ecosystem of the future. These sensing devices are connected to the internet for information exchange. The location and positioning of these nodes is very important information required in vast range of location based services like smart homes , smart healthcare , environmental monitoring, personal navigation and smart transportation. This paper presents an intelligent solution for node localization in a 6G enabled IoT network. An indoor communication network scenario is proposed in which reconfigurable intelligent surfaces (RISs) are installed to locate the sensor nodes operating in that network. The performance evaluation of the proposed scheme is carried out with optimum number of reflecting elements and optimum phase shifts. It is observed that optimized RISs with 100 reflecting elements improve the estimated localization error by 7.4% over non-optimum RISs. Also, the minimum gain of 6% in localization error is offered using equal phase shifts over random phase shifts. Further, the effect of channel conditions on the average estimation error in node locations is also elaborated. In the end, the explainable artificial intelligence (XAI) empowered indoor localization is discussed as a use case scenario and the performance comparison of the algorithms is evaluated.

Abstract With oceanic reverberation and a large amount of data being the main sources of interference for underwater acoustic target detection, it is difficult to obtain a more robust detection performance by relying on the traditional constant false alarm rate (CFAR) detection method. An adaptive sonar CFAR detection method based on a back propagation (BP) neural network is proposed. The method combines the artificial intelligence algorithm and the traditional detection algorithm, and uses the classification ability of the algorithm to select the detection algorithm, which can effectively improve the adaptation ability of the algorithm and the environment and the false alarm control ability. The method combines the artificial intelligence algorithm and the traditional detection algorithm, and uses the classification ability of the algorithm to select the detection algorithm, which can effectively improve the adaptation ability of the algorithm and the environment and the false alarm control ability. This method uses a BP neural network to train the target echo signal to complete the clutter background classification and establish the clutter background recognition classification set. According to the output result of each classification, the best CFAR detector is selected from four CA/SO/GO/OS‐CFAR detectors to detect the target. The simulation results show the detection performance of the proposed method in a uniform environment, a multi‐target environment, and a clutter edge environment. The results show that the environment adaptability is strong for different clutter backgrounds, which further improves the control ability of false alarms under a non‐uniform background.

With the development of the Internet of Things (IoT) , a large-scale, heterogeneous, and dynamic distributed network has been formed among IoT devices. There is an extreme need to establish a trust mechanism between devices, and blockchain can provide a zero-trust security framework for IoT. However, the efficiency of the blockchain is far from meeting the application requirements of the IoT, which has become the biggest resistance to the application of the blockchain in the IoT. Therefore, this paper combines sharding to build an effective Blockchain-based IoT data storage scheme (S-BDS) . Sharding can solve the problem of blockchain capacity and scalability. While the blockchain provides data immutability and traceability for the IoT, it also brings huge demands for data credibility verification. The communication delay in the IoT system seriously affects the security of the system, while the Merkle proof of traditional blockchain occupies a lot of communication resources. This paper constructs Insertable Vector Commitment (IVC) in the bilinear group and replaces the Merkle tree with IVC to store IoT data in the blockchain. The construct has small-sized proof. It also has the ability to record the number of updates, which can prevent replay-attacks. Experiments show that each block processes 1,000 transactions, the proof size of a single data piece is 30% of the original scheme, and proofs from different shards can be aggregated. IVC can effectively reduce communication congestion and improve the stability and security of the IoT system.

In wireless sensor networks (WSNs), data aggregation (DA) has become one of the most practical techniques to reduce processing delay and improve energy efficiency. To support intelligent applications, sensor nodes need to report heterogeneous and diverse data, which induce the demand for multidimensional DA and multifunctional data analysis. To solve the current security problems and functional requirements, we propose a multifunctional and multidimensional secure DA scheme to strike the balance between data availability and privacy. First, we design a Chinese remainder theorem conversion method with the counter to encode multidimensional data into large integers, which can be operated by linear homomorphic encryption schemes. Then, we introduce a multifunctional data analysis method supporting diversified aggregation functions, including linear, polynomial, and continuous functions. Moreover, we demonstrate that the proposed scheme can achieve confidentiality, integrity, authentication, and resistance against false data injection attacks. The experimental results show that the supported max dimension of one ciphertext in our scheme is at least twice that of existing schemes. Thus, in scenarios with high dimensions, our scheme is superior to the existing schemes in terms of computation and communication costs.

The next generation of wireless communication systems will rely on advantageous sixth-generation wireless network (6G) features and sophisticated edge Internet-of-Things technology to provide continuous service delegation and resource allocation. Network slicing and virtualization are common in these scenarios to meet user demands and application services. This article introduces a Normalized Slicing-assigned Virtualization Method for satisfying the 6G features in future generation systems. The proposed method relies on available resource roots and time intervals for replications. Based on the availability and Accessibility, the resource virtualization and network slicing processes are forwarded. The proposed method exploits federated learning for determining availability and accessibility models in detecting slicing, virtualization, or both the requirements throughout the resource sharing process. This improves the resource sharing rate, with less latency and high processing despite the user and application demands. The learning models are trained to balance replication and network slicing for confining complexity across different resources. The proposed method's performance is validated using the above metrics for varying users and intervals.

This paper addresses the issues of selfishness, limited network resources, and their adverse effects on real-time dissemination of Emergency Warning Messages (EWMs) in modern Autonomous Moving Platforms (AMPs) such as Vehicular Social Networks (VSNs). For this purpose, we propose a social intelligence based identification mechanism to differentiate between a selfish and a cooperative node in the network. Therefore, we devise a crowdsensing based mechanism to calculate a tie-strength value based on several social metrics. Moreover, we design a recursive evolutionary algorithm for each node’s reputation calculation and update. Given that, then we estimate each node’s state-transition probability to select a super-spreader for rapid dissemination. In order to ensure a seamless and reliable dissemination process, we incorporate 5G network structure instead of conventional short range communication which is used in most vehicular networks at present. Finally, we design a real-time dissemination algorithm for EWMs and evaluate its performance in terms of network parameters such as delivery-ratio, delay, hop-count, and message-overhead for varying values of vehicular density, speed, and selfish nodes’ density based on realistic vehicular mobility traces. In addition, we present a comparative analysis of the performance of the proposed scheme with state-of-the-art dissemination schemes in VSNs.

The car-following model describes the microscopic behavior of the vehicle. However, the existing car-following models set the drivers’ reaction time to a fixed value without considering its dynamics. In order to improve the accuracy of car-following model, this paper proposes Deep Feature Learning-based Car-Following Model (DeepCF), a car-following model based on fatigue driving and Generative Adversarial Networks (GAN). The model is composed of the drivers’ reaction time model and the car-following decision algorithm. First, we regard driving fatigue as the starting point to study the influence of driving time and the acceleration of the preceding vehicle on the drivers’ reaction time, and develop a coarse-grained drivers’ reaction time model. Secondly, considering the impact of fatigue driving on car-following decisions, we utilize GAN to generate a driving decision database based on reaction time and use Euclidean distance as a decision search indicator. Finally, we conduct experiments on a real data set, and the results indicate that our DeepCF model is superior to baseline models.

In this paper, we introduce a new fuzzy c-means (FCM) method in order to improve the magnetic resonance images’ (MRIs) segmentation. The proposed method combines the FCM and possiblistic c-means (PCM) functions using a weighted Gaussian function. The weighted Gaussian function is given to indicate the spatial influence of the neighbouring pixels on the central pixel. The parameters of weighting coefficients are automatically determined in the implementation using the Gaussian function for every pixel in the image. The proposed method is realised by modifying the objective function of the PCM algorithm to produce memberships and possibilities simultaneously, along with the usual point prototypes or cluster centres for each cluster. The membership values can be interpreted as degrees of possibility of the points belonging to the classes, that is, the compatibilities of the points with the class prototypes to overcome the coincident clusters problem of PCM. The efficiency of the proposed algorithm is demonstrated by extensive segmentation experiments using MRIs and comparison with other state-of-the-art algorithms. In the proposed method, the effect of noise is controlled by incorporating the possibility (typicality) function in addition to the membership function. Consideration of these constraints can greatly control the noise in the image as shown in our experiments.